# Partial Fractions

Partial fractions is the opposite of adding fractions over a common denominator. It applies to integrals of the form

The idea is to break into a sum of smaller terms which are easier to integrate.

(A function of the form , where and are polynomials, is called a rational function.)

I'll start by doing an example to give you a feel for the procedure. Then I'll go back and explain the steps in the method. The procedure is a bit long, and requires a substantial amount of algebra. Therefore, before using partial fractions, you should be sure that there isn't an easier way to do the integral.

First, I want to mention a formula that often comes up in these problems:

(Do you see how to work it out? Substitute , so .) For example,

Example. Compute .

. Write

Multiply both sides by to clear denominators:

Let . I get

Let . I get

Therefore,

So

Now I'll describe the steps in the method. Some of this will seem a little abstract until you see some examples.

Consider an integral of the form

where and are polynomials.

Recall that the degree of a polynomial is the highest power of the variable that occurs in it. Nonzero constants have degree 0; by convention, 0 has degree .

Step 1. If the degree of the top is greater than or equal to the degree of the bottom, divide the bottom into the top.

Step 2. You will now have an integral that looks like

where and are polynomials, and the top is smaller in degree than the bottom.

Factor the bottom of the fraction into a product of linear terms and irreducible quadratic terms.

• A linear term is a term where the variable occurs only to the first power. Here are some linear terms:

• An irreducible quadratic term is a quadratic term with only imaginary (complex) roots. That is, it is a quadratic which "doesn't factor". Here are some irreducible quadratic terms:

You can check that a quadratic is irreducible by using the general quadratic formula to find its roots. If the roots are complex numbers, the quadratic does not factor.

Warning: is not irreducible:

("Ugly" factors are allowed.) And is not considered an irreducible quadratic: It is , the square of a linear term. This distinction will become important in Step 3.

Step 3. Obtain the partial fractions decomposition for the fraction.

This is the heart of the partial fractions method. It is basically a lot of algebra, but it's sufficiently complicated that the best way to describe it is by doing some examples.

Example. Compute .

The top has degree 3 while the bottom has degree 2. Divide the bottom into the top:

So

Integrate by substitution:

Hence,

In this problem, division simplified the integral enough that it wasn't necessary to go any further in the partial fractions procedure.

Example. Consider the integral . The top and the bottom have the same degree. Divide the bottom into the top:

Therefore,

I can do easily; I need to integrate the second term. Factor the denominator: . Then

Let . I get , so .

Let . I get , so .

Thus,

Therefore,

Long division in Step 1 is a preliminary operation which puts the integral into the right form for the rest of the procedure. If you do a division, check before going on to see whether you can use a simple technique (like substitution) to do the integrals you've obtained. Sometimes you can complete the integration immediately; otherwise, you'll need to go on to Step 2.

Example. Compute .

This example will show how to handle repeated factors --- in this case, . Here's what you do:

For a repeated factor, you have one term for each power up to the power the factor is raised to. In this case, you have a term for and a term for .

Multiply to clear denominators:

Let . I get .

Let . I get .

Plug the a and c values back in:

With only b left, I can plug in any number and solve for b. I'll let :

Therefore,

Hence,

Alternatively, take equation (*). Multiply out the b-term:

Since this is an identity, I can differentiate both sides:

Now I can recycle . Plugging in gives , so as before.

At any point, you can plug in an arbitrary number for x, or you can differentiate both sides of the equation.

Example. Compute .

In this case, the repeated factor is the " ". I want a, b, and c so that

Clear denominators:

Let . I obtain , so .

Next, let . Then so .

Put and back into the equation:

Let . I get , so .

Substitute the a, b, and c values into the original decomposition:

Finally, do the integral:

Example. How would you try to decompose

using partial fractions? That is, what is the initial partial fractions equation?

The linear factor is repeated 4 times, and the linear factor is repeated 2 times. So you use

You could do the terms first instead. Notice that the numerator has no effect on the decomposition.

Example. How would you try to decompose

using partial fractions? That is, what is the initial partial fractions equation?

The linear factor x is repeated 3 times and the linear factor is repeated twice. Therefore, you should try to solve

Notice that the top of the fraction is irrelevant in deciding how to set up the decomposition. It only comes in during the solution process.

Notice also that " " is considered a linear term (x) raised to the third power. You get one term on the right for x, one for , and one for --- no "skipping"!

Example. Compute .

The quadratic factor is irreducible. Here's how I try to decompose the fraction:

Thus, a quadratic factor (or a quadratic factor to a power) will produce terms on the right with "two letters" on top.

The rationale is the same as the one I gave for repeated factors. I don't know what kind of fraction to expect, so I have to take the most general case.

Clear denominators:

Let . I get , so . Plug back in:

Let . I get , so , or . Plug back in:

At this point, I've run out of "nice" numbers to plug in for x. There are several ways to proceed.

First, since I only have one letter to find (b), I can plug in a random value for x and solve. For example, plugging in gives . Simplifying gives , so , or .

Second, since the equation is an identity, I can {\it differentiate both sides}. Doing so, I get

(I got two b-terms by applying the Product Rule to .) Now I can recycle an old x-value, say . Plugging this in gives , or , so again.

You can use any combination of differentiating and plugging in that you wish.

Thus,

The first and third terms come from basic formulas. I integrated the second term using the substitution .

You handle repeated quadratic factors just as you handle repeated linear factors.

Example. How would you try to decompose

using partial fractions? That is, what is the initial partial fractions equation?

The quadratic is irreducible. Try the decomposition

Example. Compute .

Try the decomposition

Let . I get , so . Plug back in to obtain

Let . I get , so . Plug back in to obtain

I can differentiate both sides, or I can plug in another value of x. I'll plug in . Doing so gives , or , so .

Thus, the integral becomes

The first and third terms are integrated using basic formulas; the second term is integrated using the substitution .

Example. Compute .

Try the decomposition

In this case, there is no value of x I can plug in which will allow me to solve for one of a, b, c, d immediately. Therefore, I'll have to plug in and get equations, which I'll solve simultaneously later.

Let . This gives .

Differentiate the equation to obtain

Let . This gives .

Before I differentiate again, I'll multiply the -terms in, so that I can avoid using the Product Rule:

Differentiate:

Let . This gives , or , so . Plug this into to obtain , or . Then gives .

Differentiate one more time:

Thus, , or . Solve simultaneously with :

Therefore, . Plugging into gives .

Therefore, the integral becomes

The first and third terms are integrated using basic formulas. The second term is integrated using the substitution .

Example. Compute .

The quadratic does not factor. Try the decomposition

Let . I get , so . Plug back in to obtain

Let . I get , so . Plug back in to obtain

Differentiate:

Let . I get . so .

Therefore, the integral becomes

I'll do the second term separately, since the first term is easy. The idea is to complete the square, then do a substitution:

Hence,

Example. Compute .

First,

In this example, there's an irreducible quadratic factor . I try the decomposition

Clear denominators:

Let . Then , so . Plug it back in:

Let . I get , so . Plug it back in:

Now I can either plug in a value for x at random, or differentiate. I'll differentiate:

Differentiate again:

This gives .

Plug the values back into the original fractional decomposition:

The integral is

I'll do the integrals separately. First,

Next,

I can do the first integral using a substitution:

The second requires completing the square:

Putting the two together,

Finally, answer to the original problem is

What a mess! This is why you should consider other methods before you turn to partial fractions!